Every assessed borrower, ranked, banded, and ready for review.
| Borrower | Segment | Requested | Score | Verdict | Health card |
|---|---|---|---|---|---|
Scores are computed by a monotonic LightGBM ensemble with TreeSHAP reason codes, calibrated to a 12-month PD. Registry and phoenix overlays are deterministic and can only worsen a verdict — never improve it. Demo personas run on synthetic consented data. Want to score a business of your own? Try the live demo →